Update app.py
Browse files
app.py
CHANGED
|
@@ -29,7 +29,7 @@ model = YOLO(model_path).to(device)
|
|
| 29 |
|
| 30 |
# Define the detection function for Gradio
|
| 31 |
@spaces.GPU # Decorator to allocate GPU for ZeroGPU-enabled Spaces
|
| 32 |
-
def detect_objects(image: np.ndarray) -> Image.Image:
|
| 33 |
# Ensure the image is in BGR format if provided by PIL (Gradio gives us an RGB image)
|
| 34 |
if image.shape[-1] == 3:
|
| 35 |
image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
|
|
@@ -57,6 +57,9 @@ def detect_objects(image: np.ndarray) -> Image.Image:
|
|
| 57 |
# Apply Non-Maximum Suppression (NMS) to the detections to avoid duplicate boxes
|
| 58 |
detections = detections.with_nms(threshold=NMS_THRESHOLD, class_agnostic=False)
|
| 59 |
|
|
|
|
|
|
|
|
|
|
| 60 |
# Initialize an annotator for bounding boxes with specified color and thickness
|
| 61 |
box_annotator = sv.OrientedBoxAnnotator(color=ANNOTATION_COLOR, thickness=ANNOTATION_THICKNESS)
|
| 62 |
|
|
@@ -65,11 +68,13 @@ def detect_objects(image: np.ndarray) -> Image.Image:
|
|
| 65 |
|
| 66 |
# Convert annotated image to RGB for Gradio display (PIL expects RGB)
|
| 67 |
annotated_img_rgb = cv2.cvtColor(annotated_img, cv2.COLOR_BGR2RGB)
|
| 68 |
-
|
|
|
|
|
|
|
| 69 |
|
| 70 |
# Reset function for Gradio UI
|
| 71 |
def gradio_reset():
|
| 72 |
-
return gr.update(value=None), gr.update(value=None)
|
| 73 |
|
| 74 |
# Set up Gradio interface
|
| 75 |
with gr.Blocks() as demo:
|
|
@@ -83,6 +88,7 @@ with gr.Blocks() as demo:
|
|
| 83 |
|
| 84 |
with gr.Column():
|
| 85 |
output_img = gr.Image(label="Detection Result", interactive=False)
|
|
|
|
| 86 |
|
| 87 |
# Add Examples section with images from the root directory
|
| 88 |
with gr.Accordion("Select an Example Image"):
|
|
@@ -94,8 +100,8 @@ with gr.Blocks() as demo:
|
|
| 94 |
)
|
| 95 |
|
| 96 |
# Define button actions
|
| 97 |
-
clear.click(gradio_reset, inputs=None, outputs=[input_img, output_img])
|
| 98 |
-
predict.click(detect_objects, inputs=[input_img], outputs=[output_img])
|
| 99 |
|
| 100 |
# Launch Gradio app
|
| 101 |
demo.launch(server_name="0.0.0.0", server_port=7860, debug=True)
|
|
|
|
| 29 |
|
| 30 |
# Define the detection function for Gradio
|
| 31 |
@spaces.GPU # Decorator to allocate GPU for ZeroGPU-enabled Spaces
|
| 32 |
+
def detect_objects(image: np.ndarray) -> (Image.Image, str):
|
| 33 |
# Ensure the image is in BGR format if provided by PIL (Gradio gives us an RGB image)
|
| 34 |
if image.shape[-1] == 3:
|
| 35 |
image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
|
|
|
|
| 57 |
# Apply Non-Maximum Suppression (NMS) to the detections to avoid duplicate boxes
|
| 58 |
detections = detections.with_nms(threshold=NMS_THRESHOLD, class_agnostic=False)
|
| 59 |
|
| 60 |
+
# Count total detections after NMS
|
| 61 |
+
total_detections = len(detections)
|
| 62 |
+
|
| 63 |
# Initialize an annotator for bounding boxes with specified color and thickness
|
| 64 |
box_annotator = sv.OrientedBoxAnnotator(color=ANNOTATION_COLOR, thickness=ANNOTATION_THICKNESS)
|
| 65 |
|
|
|
|
| 68 |
|
| 69 |
# Convert annotated image to RGB for Gradio display (PIL expects RGB)
|
| 70 |
annotated_img_rgb = cv2.cvtColor(annotated_img, cv2.COLOR_BGR2RGB)
|
| 71 |
+
|
| 72 |
+
# Return the annotated image and the total count of detections
|
| 73 |
+
return Image.fromarray(annotated_img_rgb), f"Total Detections: {total_detections}"
|
| 74 |
|
| 75 |
# Reset function for Gradio UI
|
| 76 |
def gradio_reset():
|
| 77 |
+
return gr.update(value=None), gr.update(value=None), gr.update(value="")
|
| 78 |
|
| 79 |
# Set up Gradio interface
|
| 80 |
with gr.Blocks() as demo:
|
|
|
|
| 88 |
|
| 89 |
with gr.Column():
|
| 90 |
output_img = gr.Image(label="Detection Result", interactive=False)
|
| 91 |
+
detection_count = gr.Textbox(label="Detection Summary", interactive=False)
|
| 92 |
|
| 93 |
# Add Examples section with images from the root directory
|
| 94 |
with gr.Accordion("Select an Example Image"):
|
|
|
|
| 100 |
)
|
| 101 |
|
| 102 |
# Define button actions
|
| 103 |
+
clear.click(gradio_reset, inputs=None, outputs=[input_img, output_img, detection_count])
|
| 104 |
+
predict.click(detect_objects, inputs=[input_img], outputs=[output_img, detection_count])
|
| 105 |
|
| 106 |
# Launch Gradio app
|
| 107 |
demo.launch(server_name="0.0.0.0", server_port=7860, debug=True)
|